Formal infrastructure for modelling intelligent agents with agent UML and Petri nets

Abstract

The complexity of today's software systems demands specific approach to software development. While Unified Modelling Language (UML) has become standard notation for analysis and design of software systems, and its extension Agent UML (AUML) is yet to become a standard, current (A)UML specifications have many limitations because they are initially intended to be general-purpose and suitable for many different kinds of applications. On the other hand, Petri nets are a formal concept suitable for simulation, validation and verification of software system execution. This thesis is focused on analysing the current AUML specifications and the concepts of Petri nets in order to find connecting points between them and to propose the means for extending AUML with Petri nets. In order to practically examine the possibilities for extending AUML with the concept of Petri nets this thesis applied the proposed methodology for extending AUML with Petri nets to development of multi-agent system and uses AUML as a main modelling tool for multi-agent system specification and design, and Petri nets as a formal verification and validation tool before actual implementation. © Springer-Verlag 2004.